Running head: ACE GAMBLING-SUBSTANCE USE MODELING 1

Genetic and Environmental Influences on Gambling

and Substance Use in Early Adolescence

Frank Vitaro

University of Montreal

Amy C. Hartl

Florida Atlantic University

Mara Brendgen

University of Quebec at Montreal

Brett Laursen

Florida Atlantic University

Ginette Dionne and Michel Boivin

Laval University

This research was made possible by grants from the Fonds de Recherche du Québec (Société et

Culture), the Quebec Ministry of Health and Social Services, the U.S. National Institute of Child

Health and Human Development (HD068421) and the U.S. National Science Foundation

(0909733). Correspondence to: Frank Vitaro, Research Unit on Children’s Psychosocial

Maladjustment, 3050 Edouard-Montpetit Blvd., Montreal (Quebec) Canada, H3T 1J7. Email:

frank.vitaro@umontreal.ca
ACE GAMBLING-SUBSTANCE USE MODELING 2

Abstract

This study examined the genetic and environmental architecture of early gambling involvement

and substance use to determine whether genetic or environmental factors that contribute to

substance use also put young adolescents at risk for early involvement in gambling. Self-reports

of substance use and gambling involvement were collected at age 13 years from 279

Monozygotic and Dizygotic twin pairs. Univariate ACE modeling revealed that genetic and

nonshared environmental factors almost equally accounted for gambling involvement, with no

contribution from shared environmental factors. In contrast, both shared and nonshared

environmental factors played important roles in substance use; the contribution of genetic factors

was also substantial. Bivariate analyses identified a significant, albeit modest, overlap between

the genetic influence on gambling involvement and the genetic influence on substance use. The

results shed light on the etiology of early gambling involvement and substance use, suggesting

that preventive interventions targeting common risk factors may also need to be complemented

by modules that are specific to each behavior.

Keywords: ACE modeling, gambling, substance use, early adolescence

ACE GAMBLING-SUBSTANCE USE MODELING 3

Genetic and Environmental Influences on Gambling and Substance Use

in Early Adolescence

In many Western countries, one 13 year-old out of five has used drugs or alcohol during

the past year and at least as many report participation in gambling activities (Johnston et al.

2013; Volberg et al. 2011). Substance use and gambling tend to co-occur throughout

adolescence. Common risk factors inherent to the child (e.g., impulsivity) and common risk

factors related to the child’s social environment (e.g., affiliation with delinquent peers) have been

found to contribute, at least partially, to the co-occurrence of gambling and substance use in

adolescence (Barnes et al. 1999; Vitaro et al. 2001). The exact source of these common

etiological factors, however, remains unknown. Impulsivity is moderately heritable, but

environmental factors are also an important source of influence (Bezdjian et al. 2011).

Environmental features, such as affiliation with delinquent peers, may also be influenced by

genetic factors via a gene-environment correlation (Button et al. 2007). As a consequence, it is

unclear the extent to which the co-occurrence of adolescent gambling and substance use is due to

common underlying personal, heritable characteristics of the child or to common underlying

influences from the child’s environment. We address this issue with data from a genetically

informed study of monozygotic and dizygotic adolescent twins raised together.

A number of studies indicate that shared environmental factors play an important role

with respect to substance use in adolescence, with the rest of the variance explained by genetic

and nonshared environmental factors (Hopfer et al. 2003; Kendler et al. 2008). Less is known

about early gambling involvement (i.e., frequency of gambling behavior) and the evidence that is

available during adolescence is limited and contradictory. The only study examining the gene-

environment architecture of the frequency of adolescent gambling involvement found evidence

ACE GAMBLING-SUBSTANCE USE MODELING 4

of considerable shared environmental influences, with minimal genetic contribution (Loehlin and

Nichols 1976). Worth noting, however, is the fact that these data were collected more than 50

years ago, when gambling was less popular and less accepted than it is now. The other studies to

examine the gene-environment architecture of gambling focused on adults. Two reports found

that genetic factors played a significant role in gambling involvement (Slutske et al. 2009;

Winters and Rich 1998), but a third found no evidence of genetic influence (Slutske and

Richmond-Rakerd 2014). Instead, the latter found that gambling involvement was influenced by

family (i.e., shared) and non-shared environmental factors. None of these studies reported sex

differences.

To better understand the etiology of gambling involvement and substance use, a first

important question that needs to be addressed is whether the relative contributions of genetic,

shared and nonshared environmental factors to early adolescent gambling involvement are

similar to those found for early substance use. A second, equally important question is whether

the same genetic, shared and nonshared environmental factors that influence early gambling

involvement also influence early substance use. There is some empirical support for the assertion

that the same genetic factors and, to a lesser degree, the same environmental factors are involved

in gambling and substance use. This support, however, is from adult studies and refers to

gambling problems and substance\alcohol abuse and dependence (Slutske et al. 2000; Slutske et

al. 2013). The degree to which these findings generalize to the initiation of gambling and

substance use during early adolescence is not clear.

The answers to these questions have important theoretical and clinical consequences.

First, knowledge about the role of genetic and environmental influences on gambling

involvement (and substance use) during early adolescence, when most youth first engage in these
ACE GAMBLING-SUBSTANCE USE MODELING 5

behaviors will help to clarify initiation pathways. Second, delineation of the origin of the overlap

between early adolescent gambling and substance use will provide a clearer insight into the

common etiology of these two behaviors than previous studies of adults, because the phenotypic

association between gambling and substance use at initiation is relatively uncontaminated by

snowball effects resulting from mutual influences between the two behaviors over time (Wanner

et al. 2009). Finally, a better understanding of the onset of these problem behaviors will inform

prevention programs, which should help avoid adjustment problems downstream (DuRant et al.

1999; Rahman et al. 2012).

For example, if genetic factors are important, then heritable phenotypes known to predict

gambling and substance use, such as impulsivity, could be targeted. If shared environmental

factors turn out to be important, then family factors can be targeted. Unique social experiences

(i.e., peers) should be considered if nonshared environmental influences are important. Finally, if

common genetic or environmental factors are involved with both types of problem behaviors,

then prevention programs could adopt a generic perspective by targeting common risk factors

shared by both types of problem behaviors, as has been proposed elsewhere (Derevensky et al.

2004; Potenza 2003).

To address these issues, the first objective of the present study was to model the genetic

and (shared and nonshared) environmental influences on gambling involvement and on

substance use during early adolescence. The second objective of the present study was to

examine the extent to which gambling involvement and substance use during early adolescence

can be explained by the same underlying genetic, shared, and nonshared environmental factors.

Method

Participants
ACE GAMBLING-SUBSTANCE USE MODELING 6

The sample is part of an ongoing longitudinal study of a population-based sample of

twins drawn from the Quebec Newborn Twin Registry of all twin births occurring in the

Province of Quebec, Canada, between 1996 and 1998. A total of 650 families agreed to

participate in the study (113 MZ female pairs, 110 MZ male pairs, 115 DZ female pairs, 107 DZ

male pairs, 205 mixed-sex DZ twin pairs). Zygosity was assessed by genetic marker analysis of

8-10 highly polymorphous genetic markers and twins were diagnosed as MZ when concordant

for every genetic marker. When genetic material was insufficient or unavailable due to parental

refusal (43% of cases), zygosity was determined based on physical resemblance questionnaires at

18 months and again at age 9 (Goldsmith 1991). The comparison of zygosity based on

genotyping with zygosity based on physical resemblance in a subsample of 237 same-sex pairs

revealed a 94% correspondence rate, which is similar to rates obtained in other studies

(Magnusson et al. 2013; Spitz et al. 1996). Eighty-four percent (n = 546) of the families were of

European descent, 3% (n = 19) were of African descent, 2% (n = 13) were of Asian descent, 2%

(n = 13) were Native North Americans, and 9% (n = 59) did not specify ethnicity. At 5 months of

age, the demographic characteristics of the twin families resembled that of a representative

sample of singleton families from large urban centers in the province of Quebec (Santé Québec

et al. 1998). At the outset, 95% of parents lived together; 44% of the twins were the first born

children; 66% of mothers and 60% of fathers were between 25 and 34 years old; 17% of mothers

and 14% of fathers had not finished high school; 28% of mothers and 27% of fathers held a

university degree; 83% of the parents were employed; 10% of the families received social

welfare or unemployment insurance.

Data were collected every year between the ages of 1 and 13. The present study involves

the first assessment of gambling and substance use, which took place at age 13 (between 2009
ACE GAMBLING-SUBSTANCE USE MODELING 7

and 2011). Data collection at this wave was approved by the Sainte-Justine Hospital Research

Centre ethics committee. Parent consent and child assent for participation was obtained.

Of the 650 pairs in the initial sample, 401 (61.7%) participated in data collection at age

13. Same- and mixed-sex DZ twins displayed differences in the multivariate covariance matrix

(χ2(6) = 87.94, p < .001), so the 122 mixed-sex DZ twin pairs were excluded from the analyses.

The final sample included 279 twin pairs (n = 160 MZ and n = 119 same-sex DZ). The final

sample did not differ significantly from the 207 MZ and same-sex DZ twin pairs that did not

participate in the data collection at age 13 in initial family income (t(299) = 0.17, p = .86), family

structure (χ2(2) = 0.34, p = .84), or birth weight (t(840) = 0.20, p = .84). Those that did and did

not participate in the age 13 data collection also did not differ significantly on a variety of

problem behaviors, including disruptiveness (K: t(551) = 0.71, p = .48; G1: t(604) = -0.18, p

= .86), proactive aggression (K: t(551) = 1.15, p = .25; G1: t(603) = 0.29, p = .77), physical

aggression (K: t(551) = -0.13, p = .89; G1: t(604) = -0.66, p = 51.), indirect aggression (K: t(550)

= 1.13, p = .26; G1: t(604) = -0.34, p = .73), and reactive aggression (K: t(551) = 0.38, p = .70; 1:

t(604) = -0.60, p = .55) at age 6 or age 7 years, nor delinquency at age 10 years (t(610) = -1.22, p

= .22). Of the 279 twin pairs participating in the age 13 data collection, missing data accounted

for an average of 1.6% of reports (range: 1.3%-1.8%). Missing data were handled with full

information maximum-likelihood estimation (FIML), which allowed participants with

incomplete data to be included in the models. Little’s test indicated that data were missing

completely at random, χ²(2) = 0.37, p = .83.

Measures

Instruments were administered either in English (21%) or in French (79%), depending on

the language spoken by the children and their parents. Back-translation procedures were
ACE GAMBLING-SUBSTANCE USE MODELING 8

employed and bilingual translators verified the semantic similarity between the back-translated

items and the original items in the questionnaire.

Gambling Involvement. Frequency of involvement in gambling was assessed with the

South Oaks Gambling Screen for adolescents (SOGS-RA) (Winters et al. 1993). Involvement

was computed by averaging the frequency scores of 12 gambling activities over the past 12-

months (e.g., purchased lottery tickets (6/49 or Banco), played Mise-O-Jeu (online sports betting

game), bought scratch offs, played Bingo for money, bet on games on the internet, played video

lottery terminal games, played cards or games with others for money, bet on sporting

events/games for money, gambled at a casino, bet on games of skill (pool, basketball), played

dice games for money, and bet on other games). Participants rated the frequency of each activity

on a scale ranging from 1 (never) to 7 (daily) (Cronbach’s α = .65). MZ and DZ twins had

identical average scores (M = 1.03; MZ SD = 0.09; DZ SD = 0.10). Approximately 20% of

participants engaged in at least one gambling activity at least once. This proportion is in line with

other prevalence studies (Pica et al. 2012; Volberg et al. 2011). Approximately 1.1% of

participants engaged in at least one gambling activity regularly (i.e., more than once a month).

Given the positively skewed distribution of gambling involvement, the negative reciprocal of the

measure was calculated and then z-standardized within sex, normalizing the measure (skew

before and after transformation = 7.20 and 3.16).

Substance Use. Frequency of substance use was assessed with the Personal Experience

Screening Questionnaire (PESQ) (Henly and Winters 1989; Winters et al. 1990-91). Substance

use was computed by averaging the standardized scores for four items (frequency of alcohol,

marijuana, and other drug use and binge drinking). Participants rated the frequency of substance

use over the past 12 months on a scale ranging from 1 (never) to 7 (daily), with the exception of
ACE GAMBLING-SUBSTANCE USE MODELING 9

binge drinking, which was rated on a scale from 1 (never) to 6 (5 times or more) (Cronbach’s α =

.74). MZ and DZ twins had near identical average scores (MZ M = 1.14, SD = 0.25; DZ M =

1.15, SD = 0.36). Approximately 21% of participants reported experience with at least one

substance at least once. This proportion is also in line with past prevalence studies (Johnston et

al. 2013; Pica et al. 2012). Approximately 2.1% of participants used at least one substance

regularly (i.e., more than once a month). Given the positively skewed distribution of substance

use, the negative reciprocal of the measure was calculated and then z-standardized within sex,

normalizing the measure (skew before and after transformation = 10.20 and 2.10). To take into

account the non-normality of the data in reference to both substance use and gambling

involvement, we used maximum likelihood estimation with robust standard errors (MLR).

Plan of Analysis

Univariate Genetic Models. The data were analyzed using Mplus v7.0 (Muthén and

Muthén 1998-2013). Univariate models were fit to the data to estimate the relative contribution

of genetic and environmental factors to substance use and gambling involvement. By comparing

within-pair correlations for MZ twins (who are genetically identical) and DZ twins (who on

average share only half of their genes), sources of variability in a measured variable (phenotype)

can be estimated as latent additive genetic (A), latent shared environmental (C), and latent

nonshared environmental (E) factors (Neale and Cardon 1992).

Within-twin pair correlations of the latent genetic factors (A) were fixed to 1.0 for MZ

twins and to 0.5 for DZ twins. Within-twin pair correlations of the latent shared environmental

factors (C) were fixed to 1.0 for both MZ and DZ twins. Within-twin pair correlations of the

latent nonshared environmental factors (E) were fixed to 0 for both MZ and DZ twins. The

estimated coefficients a, c, and e were fixed to be equal across the two members of a twin pair
ACE GAMBLING-SUBSTANCE USE MODELING 10

and across MZ and DZ twins. These coefficients are factor loadings that provide information

about the relative contribution of the latent factors A, C, and E to the total variance V T of each

phenotype (VT = a2 + c2 + e2). Measurement error is included in e2.

A model estimating a possible dominant genetic effect (Neale and Cardon 1992) would

also be modeled if the within-pair correlation for MZ twins is more than twice as strong as the

within-pair correlation for DZ-twins. Latent factors are removed from the models if model fit

improves or if parsimony improves without sacrificing model fit. Univariate models were chosen

on the basis of parsimony, low and nonsignificant χ2 values, and values of RMSEA below .08.

Bivariate Genetic Models. The bivariate model combines the two final best-fitting

univariate models. Bivariate genetic models are used to examine whether the phenotypic

association between substance use and gambling involvement can be explained by common

underlying factors. A bivariate Cholesky model was specified such that the covariance structure

of substance use and gambling involvement was partitioned into (1) common latent genetic,

shared and nonshared environmental factors that influence both substance use and gambling, and

(2) unique latent genetic, shared and nonshared environmental factors that are specific to

gambling. A significant contribution of a common latent factor (genetic, shared or nonshared

environmental) to both phenotypes indicates an overlap or correlation between substance use and

gambling involvement. Overlap was only estimated for latent factors that were statistically

significant in both univariate models. Latent factor effects that were estimated to be zero were

removed to maximize model parsimony and statistical power. The final bivariate results

represent the best-fitting, most parsimonious model.

A multiple group framework was used to examine sex differences in the final bivariate

model. Effects for boys and girls were constrained to be equal. There were no statistically
ACE GAMBLING-SUBSTANCE USE MODELING 11

significant changes in model fit, indicating that the pattern of results did not differ as a function

of sex.

Results

Preliminary Analyses

Two sets of nested χ2-difference tests examined mean level differences in substance use

and gambling involvement. Sex differences emerged for gambling involvement, χ2(1) = 12.74, p

< .001. Boys reported more gambling involvement than girls. There were no sex differences in

substance use, χ2(1) = 0.48, p = .49.To control for the mean level differences, scores were z-

standardized by sex, separately for MZ and DZ pairs (see Arseneault et al. 2003; Van den Oord

et al. 2000). There were no differences between MZ and DZ twins on substance use, χ2(1) =

0.002, p = .96, or gambling involvement, χ2(1) = 0.01, p = .92.

Statistically significant within-pair correlations emerged for MZ twin pairs and DZ twin

pairs on substance use (MZ r = .54, DZ r = .39) and gambling involvement (MZ r = .55, DZ r

= .18). Correlation contrasts revealed that the within-pair correlation on gambling involvement

was stronger for MZ twin pairs than for DZ twin pairs (p < .001). There were no statistically

significant differences between MZ twin pairs and DZ twin pairs in the within-pair correlation on

substance use. There were no sex differences on the within-pair correlations for substance use

(male r = .41, female r = .52) or gambling involvement (male r = .33, female r = .46). The

phenotypic correlation between substance use and gambling involvement was r = .21, p < .001

(CI = 0.12; 0.29).

Univariate Genetic Models

Table 1 summarizes the results of the univariate models for substance use and for

gambling involvement. For substance use, genetic factors explained about 33% of the variance,
ACE GAMBLING-SUBSTANCE USE MODELING 12

shared environmental sources explained about 22% of the variance, and nonshared

environmental factors accounted for 45% of the variance. For gambling involvement, genetic

factors and nonshared environmental factors each accounted for about half of the variance; the

effect of shared environmental factors was estimated to be 0. When the shared environmental

effects were fixed to zero in an AE model of gambling involvement, there was no statistically

significant change in model fit, χ2(1) = .00, p = 1.00. The within-pair correlation for MZ twins

was more than twice as strong as the within-pair correlation for DZ-twins, so we also estimated

an ADE model with a dominant genetic effect for gambling involvement (Neale and Cardon

1992). The AE model provided a better fit to the data than the ADE or the ACE model.

Bivariate Genetic Models

The bivariate Cholesky model based on the final univariate models included two

components: (1) common latent factors AC, CC, and EC that influence both phenotypes, and (2)

unique latent factors AUG, CUG, and EUG specific to gambling involvement (see Figure 1). The

shared environmental effect on gambling involvement was not statistically significant in

univariate analyses, so the C component (cUG) and C overlap (cCG) parameters were fixed to zero

in the bivariate model. Because the nonshared environmental overlap was estimated to be zero in

the full bivariate model, the parameter estimate of the factor loading of gambling on E C was

fixed to zero in the final bivariate model (see Table 2). There was no statistically significant

difference in model fit between the full model estimating the E overlap (i.e., e CG) and the model

constraining this coefficient to zero, χ2(1) = 0.26, p = .61. The final model (Figure 1) refers to

the most parsimonious model with the nonsignificant coefficients constrained to zero. The model

fit the data well, χ2(8) = 3.86, p = .87; RMSEA = .0.05, CFI = .96.

The results revealed that the relative contribution of the ‘common’ latent genetic factor
ACE GAMBLING-SUBSTANCE USE MODELING 13

AC to the genetic variance of gambling involvement (aCG2/(aCG2 + aUG2)) is 21.9% (.332/(.332

+ .632)). Thus, 22% of the genetic variance of gambling involvement was explained by genetic

factors that also influenced substance use. The phenotypic association between substance use and

gambling involvement was entirely explained by common underlying genetic influences; all

other covariance paths were zero.

Supplemental Analyses

Analyses were rerun with count variables for gambling involvement and substance use

comparable to those used elsewhere {e.g.`, \Slutske, 2013 #8227}. Count variables were

calculated as the number of items that participants responded to with anything other than

"never"gambling activities the participants engaged in or the number of substances the

participants used over the past 12 months. The patterns of significance were the same in the final

model with one exception: the shared environment (C) component of substance use was weaker

and the A component was stronger.

Discussion

This study represents the first twin study of gambling involvement during early

adolescence. Its aims were a) to examine the relative contribution of genetic and environmental

factors to gambling involvement and substance use during an age period (i.e., early adolescence)

when most youth first engage in these behaviors, and b) to investigate whether the concomitant

early emergence of gambling involvement and substance use can be explained by common

underlying genetic factors and common underlying environmental factors.

As expected, age 13 gambling involvement and substance use were correlated, albeit

modestly, at the phenotypic level. This correlation did not differ for males and females. Similar

correlations between gambling involvement and substance use have been found in studies of
ACE GAMBLING-SUBSTANCE USE MODELING 14

older adolescentswith adolescent samples (Barnes et al. 2005; Richmond-Rakerd et al. 2013;

Vitaro et al. 2001) and younger adolescents {Richmond-Rakerd, 2013 #8229}.

The results from our univariate ACE models for substance use are consistent with past

findings indicating that genetic, shared, and nonshared environmental factors play a significant

role in early adolescent substance use (Hopfer et al. 2003; Kendler et al. 2008). The results for

early adolescent gambling involvement, however, resemble those for gambling involvement and

problem gambling in adults (Beaver et al. 2010; Eisen et al. 1998; Slutske et al. 2009; Slutske et

al. 2010; Winters and Rich 1998), in that half or more of the variance was explained by genetic

factors and the rest by nonshared environmental factors (see, however, Slutske and Richmond-

Rakerd 2014, for an exception). The present results raise two questions. First, why did we find

genetic influences for gambling involvement in early adolescents, whereas the only other study

that examined gambling involvement in adolescents (i.e., Loehlin and Nichols 1976) did not?

One line of explanation lies in social norms. Gambling data for the Loehlin and Nichols (1976)

study were collected in an era (the early 1960s) when gambling was generally prohibited. In

contrast, the current data were collected in more permissive times. As a consequence,

adolescents in earlier studies, although older than the adolescents in the present study, may have

been less likely to express their genetic dispositions towards gambling than adolescents in the

current study. In other words, social norms may have suppressed the expression of genetic

influences through a gene x environment interaction. The existence of moderating effects of

social norms on the expression of genetic risk for other phenotypes, such as aggression, gives

credence to this position (Brendgen et al. 2013).

A similar line of explanation can help address a second question: Why do shared

environmental factors play a role with respect to substance use during early adolescence and not
ACE GAMBLING-SUBSTANCE USE MODELING 15

with respect to gambling involvement? Results from a recent survey of parents with adolescents

between the ages of 13 and 18 implicate parent attitudes. In this survey, parents indicated that

they view alcohol and drug use as a much greater problem than gambling (Campbell et al. 2011).

A general absence of parental norms or family rules that discourage gambling in most families

may thus explain why environmental factors shared by twins raised together did not significantly

account for interindividual differences in early gambling involvement in the present study. In

contrast, the likely greater variability of family rules regarding substance use – with very lenient

rules in some families and very strict rules in other families – may explain why shared

environmental factors contributed to substance use during early adolescence in the present study.

Our bivariate Cholesky models identified significant overlap in the genetic factors

associated with both substance use and gambling involvement, which entirely explained the

phenotypic correlation between these two variables. Such findings are consistent with molecular

genetic studies indicating that gambling and substance use share biological mechanisms that

encompass impulse control networks. In particular, serotonin, involved with behavioral

inhibition, and dopamine, involved with learning, motivation, and reward sensitivity, may

contribute to both types of disorders (Potenza 2008). The degree to which overlapping genetic

factors influence substance use and gambling involvement is typically greater among adults

(Slutske et al. 2000). Studies with adults also reveal overlapping nonshared environmental

factors, which we did not find in our sample of early adolescents. The correlation between

gambling and substance use is typically greater in adults than in adolescents (Pietrzak et al.

2007) and this increased correlation may be due to an escalating interplay (i.e. mutual influence)

between substance use and gambling in adulthood (Vanyukov et al. 2012). It is thus possible that

the greater overlap of genetic and environmental influences observed in adults than in our early
ACE GAMBLING-SUBSTANCE USE MODELING 16

adolescent sample arises from increased reciprocal influences between gambling and substance

use that amplify the covariance between the two behaviors over the course of development.

Further longitudinal studies are necessary to clarify this issue.

The present study is the first to examine the genetic-environmental architecture of

gambling involvement and substance use during the critical early adolescent years, which

typically encompass the onset of involvement with substances use and gambling. Nevertheless,

several limitations merit mention. First, our interest on early adolescence gave rise to a focus on

the frequency of gambling and substance use and not on substance- or gambling-related

problems, which are rare at age 13. Although early involvement in substance use and gambling is

non-normative (only one out of five adolescents is involved) and although both behaviors are

established precursors of later problems, caution should be exercised in generalizing our findings

to the etiology of addictions. Second, we note our exclusive reliance on self-reports to assess

gambling involvement and substance use. Finally, the relatively small sample size and resulting

limited statistical power may explain the absence of sex differences.

We close with three tentative conclusions. First, gambling and substance use share a

common genetic basis in early adolescence, although most of the genetic and all of the

environmental variance is unshared during this age period. In that respect, the common etiology

of gambling involvement and substance use is, at best, partial. It is nevertheless important to

examine what specific endophenotype(s), which may themselves be heritable, mediate this

common, although modest genetic overlap. The identification of such phenotypes could be useful

both for early screening of at-risk individuals and as a potential target for preventive

interventions. As indicated in the introduction, response inhibition/impulsivity could play a role

(Bezdjian et al. 2011), but other endophenotypes such as reward dependence/delayed discounting
ACE GAMBLING-SUBSTANCE USE MODELING 17

(i.e., how much a reward loses values based on its distance in time) (MacKillop 2013) or a more

general externalizing problem syndrome (Krueger et al. 2007) could be involved. Second,

prevention programs that - in accordance with a generic perspective of prevention (Derevensky

et al. 2004; Potenza 2003) - target common risk factors shared by risk taking behaviors such as

gambling and substance use may be effective in early adolescence, but only partially if they are

not complemented by modules that are specific to each type of behavior. Finally, future research

should include contextual measures such as peer group or family norms, which could play an

important role for conditioning the expression of genetic liability. The contribution of these

variables may vary across development, because the genetic-environmental architecture of both

gambling involvement and substance use could change with age as adolescents become more

autonomous and as problem behaviors become more normative or more specialized (Vrieze et

al. 2012).
ACE GAMBLING-SUBSTANCE USE MODELING 18

References

Arseneault L, Moffitt TE, Caspi A, Taylor A, Rijsdijk FV, Jaffee SR, Ablow JC, Measelle JR

(2003). Strong genetic effects on cross-situational antisocial behaviour among 5-year-old

children according to mothers, teachers, examiner-observers, and twins' self-reports. J

Child Psychol Psychiatry 44(6): 832-848

Barnes GM, Welte JW, Hoffman JH, Dintcheff BA (1999). Gambling and alcohol use among

youth: Influences of demographic, socialization, and individual factors. Addict Behav

24(6): 749-767

Barnes GM, Welte JW, Hoffman JH, Dintcheff BA (2005). Shared predictors of youthful

gambling, substance use, and delinquency. Psychol Addict Behav 19(2): 165-174

Beaver KM, Hoffman T, Shields RT, Vaughn MG, DeLisi M, Wright JP (2010). Gender

differences in genetic and environmental influences on gambling: Results from a sample

of twins from the National Longitudinal Study of Adolescent Health. Addiction 105(3):

536-542

Bezdjian S, Baker LA, Tuvblad C (2011). Genetic and environmental influences on impulsivity:

A meta-analysis of twin, family and adoption studies. Clin Psychol Rev 31(7): 1209-1223

Brendgen M, Girard A, Vitaro F, Dionne G, Boivin M (2013). Do peer group norms moderate

the expression of genetic risk for aggression? J Crim Justice 41(5): 324-330

Button TMM, Corley RP, Rhee SH, Hewitt JK, Young SE, Stallings MC (2007). Delinquent peer

affiliation and conduct problems: A twin study. J Abnorm Psychol 116(3): 554-564

Campbell C, Derevensky J, Meerkamper E, Cutajar J (2011). Parents' perceptions of adolescent

gambling: A Canadian national study. Journal of Gambling Issues 25: 36-53

ACE GAMBLING-SUBSTANCE USE MODELING 19

Derevensky JL, Gupta R, Dickinson L, Deguire A-E (2004) Prevention efforts toward reducing

gambling problems. In: Derevensky JL, Gupta R (eds) Gambling problems in youth:

Theoretical and applied perspectives. Kluwer Academic/Plenum Publishers, New York,

NY

DuRant RH, Smith JA, Kreiter SR, Krowchuk DP (1999). The relationship between early age of

onset of initial substance use and engaging in multiple health risk behaviors among

young adolescents. Arch. Pediatr. Adolesc. Med. 153(3): 286-291

Eisen SA, Lin N, Lyons MJ, Scherrer JF, Griffith K, True WR, Goldberg J, Tsuang MT (1998).

Familial influences on gambling behavior: an analysis of 3359 twin pairs. Addiction

93(9): 1375-1384

Goldsmith HH (1991). A zygosity questionnaire for young twins: A research note. Behav Genet

21(3): 257-269

Henly GA, Winters KC (1989). Development of psychosocial scales for the assessment of

adolescent involved with alcohol and drugs. Int J Addict 24(10): 973-1001

Hopfer CJ, Crowley TJ, Hewitt JK (2003). Review of twin and adoption studies of adolescent

substance use. J Am Acad Child Adolesc Psychiatry 42(6): 710-719

Johnston LD, O’Malley PM, Bachman JG, Schulenberg JE (2013) Monitoring the Future

national survey results on drug use, 1975–2012: Vol. 1. Secondary school students.

Institute for Social Research, The University of Michigan, Ann Arbor, MI

Kendler KS, Schmitt E, Aggen SH, Prescott CA (2008). Genetic and environmental influences

on alcohol, caffeine, cannabis, and nicotine use from early adolescence to middle

adulthood. Arch Gen Psychiatry 65(6): 674-682

ACE GAMBLING-SUBSTANCE USE MODELING 20

Krueger RF, Markon KE, Patrick CJ, Benning SD, Kramer MD (2007). Linking antisocial

behavior, substance use, and personality: An integrative quantitative model of the adult

externalizing spectrum. J Abnorm Psychol 116(4): 645-666

Loehlin JC, Nichols RC (1976) Heredity, Environment and Personality: A study of 850 sets of

twins. University of Texas Press, Austin, TX

MacKillop J (2013). Integrating behavioral economics and behavioral genetics: delayed reward

discounting as an endophenotype for addictive disorders. J Exp Anal Behav 99(1): 14-31

Magnusson PKE, Almqvist C, Rahman I, Ganna A, Viktorin A, Walum H, Halldner L,

Lundstrom S, Ullen F, Langstrom N, Larsson H, Nyman A, Gumpert CH, Rastam M,

Anckarsater H, Cnattingius S, Johannesson M, Ingelsson E, Klareskog L, de Faire U,

Pedersen NL, Lichtenstein P (2013). The Swedish Twin Registry: Establishment of a

Biobank and Other Recent Developments. Twin Research and Human Genetics 16(1):

317-329

Muthén LK, Muthén BO (1998-2013) MPlus User's Guide. Seventh Edition. Muthén & Muthén,

Los Angeles, CA

Neale MC, Cardon LR (1992) Methodology for genetic studies in twins and families. Kluwer

Academic Publishers, Boston, MA

Pica LA, Traoré I, Bernèche F, Laprise P, Cazale L, Camirand H, Berthelot M, Plante N (2012)

L’Enquête québécoise sur la santé des jeunes du secondaire 2010-2011 [The 2010-2011

health survey of high school students]. Institut de la statistique du Québec, Québec,

QC
ACE GAMBLING-SUBSTANCE USE MODELING 21

Pietrzak RH, Morasco BJ, Blanco C, Grant BF, Petry NM (2007). Gambling level and

psychiatric and medical disorders in older adults: Results from the national epidemiologic

survey on alcohol and related conditions. Am J Geriatr Psychiatry 15(4): 301-313

Potenza MN (2003) A perspective on adolescent gambling: Relationship to other risk behaviors

and implications for prevention strategies. In: Romer D (ed), Reducing adolescent risk:

An integrated approach. Sage Publications, Thousand Oaks, CA

Potenza MN (2008). The neurobiology of pathological gambling and drug addiction: An

overview and new findings. Philos Trans R Soc Lond B Biol Sci 363(1507): 3181-3189

Rahman AS, Pilver CE, Desai RA, Steinberg MA, Rugle L, Krishnan-Sarin S, Potenza MN

(2012). The relationship between age of gambling onset and adolescent problematic

gambling severity. Journal of Psychiatric Research 46(5): 675-683

Richmond-Rakerd LS, Slutske WS, Heath AC, Martin NG (2013). Genetic and environmental

influences on the ages of drinking and gambling initiation: Evidence for distinct

etiologies and sex differences. Alcoholism 37: 131A-131A

Santé Québec, Jetté M, Desrosiers H, Tremblay RE (1998) “In 2001…I’ll be 5 years old!”

Survey of 5-month old infants, Preliminary report of the Longitudinal Study of Child

Development in Québec. Bibliothèque Nationale du Québec

Slutske WS, Eisen S, True WR, Lyons MJ, Goldberg J, Tsuang MT (2000). Common

genetic vulnerability for pathological gambling and alcohol dependence in men. Arch

Gen Psychiatry 57(7): 666-673

Slutske WS, Ellingson JM, Richmond-Rakerd LS, Zhu G, Martin NG (2013). Shared genetic

vulnerability for disordered gambling and alcohol use disorder in men and women:
ACE GAMBLING-SUBSTANCE USE MODELING 22

Evidence from a National Community-Based Australian Twin Study. Twin Research and

Human Genetics 16(2): 525-534

Slutske WS, Meier MH, Zhu G, Statham DJ, Blaszczynski A, Martin NG (2009). The Australian

Twin Study of Gambling (OZ-GAM): Rationale, Sample Description, Predictors of

Participation, and a First Look at Sources of Individual Differences in Gambling

Involvement. Twin Research and Human Genetics 12(1): 63-78

Slutske WS, Richmond-Rakerd LS (2014). A closer look at the evidence for sex differences in

the genetic and environmental influences on gambling in the National Longitudinal Study

of Adolescent health: from disordered to ordered gambling. Addiction 109(1): 120-127

Slutske WS, Zhu G, Meier MH, Martin NG (2010). Genetic and environmental influences on

disordered gambling in men and women. Arch Gen Psychiatry 67(6): 624-630

Spitz E, Moutier R, Reed T, Busnel MC, Marchaland C, Roubertoux PL, Carlier M (1996).

Comparative diagnoses of twin zygosity by SSLP variant analysis, questionnaire, and

dermatoglyphic analysis. Behav Genet 26(1): 55-63

Van den Oord EJCG, Boomsma DI, Verhulst FC (2000). A study of genetic and environmental

effects on the co-occurrence of problem behaviors in three-year-old twins. J Abnorm

Psychol 109(3): 360-372

Vanyukov MM, Tarter RE, Kirillova GP, Kirisci L, Reynolds MD, Kreek MJ, Conway KP,

Maher BS, Iacono WG, Bierut L, Neale MC, Clark DB, Ridenour TA (2012). Common

liability to addiction and "gateway hypothesis": Theoretical, empirical and evolutionary

perspective. Drug Alcohol Depend 123: S3-S17

ACE GAMBLING-SUBSTANCE USE MODELING 23

Vitaro F, Brendgen M, Ladouceur R, Tremblay RE (2001). Gambling, delinquency, and drug use

during adolescence: Mutual influences and common risk factors. J Gambl Stud 17(3):

171-190

Volberg RA, Gupta R, Griffiths MD, Olason DT, Delfabbro P (2011) An international

perspective on youth gambling prevalence studies. In: Derevensky J, Shek D, Merrick J

(eds) Youth gambling problems: The hidden addiction. De Gruyter, Berlin, Germany

Vrieze SI, Hicks BM, Iacono WG, McGue M (2012). Decline in genetic influence on the co-

occurrence of alcohol, marijuana, and nicotine dependence symptoms from age 14 to 29.

Am J Psychiat 169(10): 1073-1081

Wanner B, Vitaro F, Carbonneau R, Tremblay RE (2009). Cross-lagged links among gambling,

substance use, and delinquency from midadolescence to young adulthood: Additive and

moderating effects of common risk factors. Psychol Addict Behav 23(1): 91-104

Winters KC, Rich T (1998). A twin study of adult gambling behavior. J Gambl Stud 14(3): 213-

225

Winters KC, Stinchfield R, Fulkerson J (1993). Patterns and characteristics of adolescent

gambling. J Gambl Stud 9(4): 371-386

Winters KC, Stinchfield RD, Henly GA, Schwartz RH (1990-91). Validity of adolescent self-

report of alcohol and other drug involvement. Int J Addict 25(11A): 1379-1395
 ACE GAMBLING-SUBSTANCE USE MODELING 24
Table 1

Univariate Model Results

RMSE
%a2 %c2 %e2 %d2 AIC BIC CFI χ2 (df) p
A
Substance Use
ACE 32.8 22.0 45.2 1506.
1491.9 0.95 0.07 0.01 (2) 0.9
(-2.2;67.8) (-8.7;52.7) (34.5;55.9) 4
AE 56.4 43.6 1502.
- 1491.6 0.94 0.07 1.77 (3) 0.6
(46.7;66.1) (33.9;53.3) 5
CE 47.5 52.5 1504.
- 1493.4 0.91 0.08 3.56 (3) 0.3
(38.1;56.8) (43.2;61.9) 3
E 100.0 1565.
- - 1558.2 0.00 0.25 70.40 (4) 0.0
(--;--) 5
Gambling Involvement
ACE 51.9 0 48.1 1527.
1513.2 1.00 0.00 1.06 (2) 0.5
(41.3;62.4) (.0;.0) (37.6;58.7) 8
AE 51.9 48.1 1522.
- 1511.2 1.00 0.00 1.06 (3) 0.7
(41.3;62.4) (37.6;58.7) 1
CE 39.6 60.4 1533.
- 1522.5 0.88 0.08 12.35 (3) 0.0
(29.6;49.6) (50.4;70.4) 4
E 100 1574.
- - 1566.7 0.07 0.22 58.56 (4) 0.0
(--;--) 0
ADE 20.9 47.0 32.1 1527.
1512.5 1.00 0.00 0.33 (2) 0.8
(-53.0;94.7) (36.6;57.4) (-43.4;107.7) 0

Note. Best fitting models are in bold.

ACE GAMBLING-SUBSTANCE USE MODELING 25

Figure 1. Bivariate Cholesky model of substance use and gambling involvement for one member
of each twin pair.

AC AUG

aCG .33*
aCS .58** (.11; .56) aUG .63**
(.28; .88) (.48; .68)

Substance Use (S) Gambling (G)

eCS.68** eCG
(.60; .75) eUG.70**
0
(.62; .77) cUG
cCS .46* EC
(.13; .79) 0
cCG EUG CUG
CC 0

Note. The model partitions variance into (1) common latent factors AC, CC, and EC that influence

both phenotypes, and (2) unique latent factors AUG, CUG, and EUG specific to gambling

involvement. Coefficients aCS, cCS, and eCS represent the factor loadings of substance use (S) on

the common latent factors AC, CC, and EC, respectively. Coefficients aCG, cCG, and eCG represent the

factor loadings of gambling involvement (G) on the common latent factors AC, CC, and EC,

respectively. Coefficients aUG, cUG, and eUG represent the factor loadings of gambling involvement

on the unique latent factors AUG, CUG, and EUG. Univariate analyses found no significant shared

environmental effect on gambling involvement, so the C component of gambling (cUG) and C

overlap (cCG) parameters were fixed to zero. The parameter estimate eCG of the factor loading of

gambling on EC was estimated to be zero in a preliminary model and therefore fixed to be zero in

the final model. Confidence intervals (95%) are shown in parentheses. N = 279 twin pairs. *p

< .05, **p < .001, two-tailed.